Studies On Dynamic Resource Management In The Wireless Systems

Dynamic resource management as a hot direction has become an indispensablepart of wireless communication systems, especially when OFDM/MIMO is used.This paper concentrates on two points: fair scheduling in the OFDM/MIMO systemsand intercarrier Interference(ICI) suppression in the Orthogonal Frequency DiversityMultiplexing (OFDM) systems.Capacity regions of Gaussian broadcast channels and multiple-access channelsare investigated and several fair scheduling algorithms are presented with appropriatepower allocation schemes. The results are extended to parallel subchannel wirelesssystems. OFDM as a special case is studied and three fair scheduling algorithms areintroduced. It is shown the average throughput of scheduled systems increases withthe user number, signal-to-noise ratio(SNR), while scheduling gain decreases whenSNR increases. It is also found that the frequency diversity within the subbands willreduce the scheduling gain and average throughput. Therefore, it is better to keep thesubband size below the correlation bandwidth of wireless channels.In the multiple-antenna systems, the impact of SNR, user number, receiverstructures, and antenna diversity on the performance of fair scheduling is studied. It isshown that one or two receiving antennas squeezes out most advantage of fairscheduling in the single-input-multiple-output(SIMO) systems. In the MISOsystems, after scheduled, the average throughput of transmit selection diversity islarger than that of space-time orthogonal block code(STOBC). In MIMO systems,the distribution of channel capacity is given when there are two transmit antennas,and performances of Maximal Likelihood(ML) receiver, Zero-Forcing(ZF)receiver and MMSE receiver are compared. It is shown that when there are enoughusers, the average throughput of ZF receivers exceeds that of ML receivers, and theMMSE receiver is better. Therefore, the suboptimal receivers are asymptoticallybetter than ML receiver in the multiuser systems.When there are two basestations, and users are allowed to communicate with anybasestation at any time, the scheduled throughput is larger than that of the STOBCsystem when SNR ranges from 0dB to 28dB and user number is 64.. When there areM basestations, throughput of every user is greatly improved.In fair scheduling algorithms, every mobile station has to feedback its channelinformation to basestations and a great fraction of uplink bandwidth is consumed. Toreduce the uplink bandwith consumption, a modified opportunistic beamformingalgorithm in the MIMO system and an distributed algorithm are proposed. It is shownthat substantial feedback reduction is achieved with little performance degradation.When the average number of feedback users is 4, the distributed algorithm incurslittle loss compared with the standard HDR scheduling algorithm.Due to the orthogonality of subcarriers, OFDM is highly sensitive tosynchronization errors and channel variation. Thus a new OFDM frame structure, aswell as the demodulation algorithm, is proposed to combat the time variation ofchannels. It is shown that the proposed signaling structure almost achieves the sameperformance at two times Doppler spread of the classical OFDM. To suppress the ICIcaused by the carrier frequency offset, a new linear ICI suppression technique ispresented. The algorithm is based on the fact that the ICI matrix is cyclic and unitary,and can be estimated from pilot subcarriers. In order to improve the spectralefficiency and reduce the complexity, a partial linear ICI suppression scheme ispresented. Analysis and simulations show that 3~5 pilots improve theSignal-to-Interference ratio(SIR) by 5 dB or more.